Sediment Dynamics and the Hydromorphology of Fluvial Systems (Proceedings of a symposium held in Dundee, UK, July 2006). IAHS Publ. 306, 2006. 127

Hydromorphological adjustment in meandering river systems and the role of flood events

JANET HOOKE Department of Geography, University of Portsmouth, Buckingham Building, Lion Terrace, Portsmouth PO1 3HE, UK [email protected]

Abstract This paper takes a selection of meandering rivers in the UK to address issues of morphological adjustment and change. The occurrence of peak flow events over recent decades is analysed for patterns and trends. Certain hydrological phases are identifiable. Changes in morphology and activity of the channels are examined through repeat field and aerial photograph mapping and some detailed observations over a period of 20 years. These show major variations, some of which appear to be related to discharge fluctuations. The differing impact of the highest floods within the period of record is demonstrated. The state of the system, particularly the degree of meander development, is of crucial importance. A seasonal difference in the impact of high flow events is also detected. Relationships between morphological changes and discharge characteristics are complex and the effects of other factors, particularly growing season and ramped effects of growth of vegetation may be significant. Key words channel changes; channel migration; discharge fluctuations; erosion; floods; hydromorphology; river meanders; temperature changes

INTRODUCTION

Understanding of the response of river channels to climate and land-use changes has become even more important under scenarios of global warming and with new legislation, particularly the EU Water Framework Directive (WFD). Sediment movement takes place almost entirely in peak flow events but these may be of varying magnitude depending on the hydrological regime and the size range of sediment available in the system. It is primarily coarse sediment, from sand size upwards, that contributes to channel morphology. Meandering rivers are mainly developed in the flood plain zones of fluvial systems. Within these zones, the main source of coarse sediment to the channel system is erosion of the banks of the flood plains themselves, with occasional contributions from valley walls. Sediment supply from tributaries is localized in input. It is generally assumed that sediment has the potential to move down through the system. However, recent analyses have suggested that movement of the coarsest material may be much more localized in some systems and that there is not necessarily high connectivity between reaches, or that it may be operational only in very extreme events. (Fryirs & Brierley, 2001; Harvey, 2002; Hooke, 2003a). The incidence and impact of flood events in meandering rivers is examined here, through evidence of channel changes at a variety of spatial and temporal scales, to assess the conditions under which erosion and deposition take place and thus hydromorphological responses are brought about. The 128 Janet Hooke

analysis focuses on the Rivers Bollin and Dane in northwest England (Hooke 1987, 1995a,b) with additional examples from the River Severn in Wales (Hooke et al., 1994), and the River Wey in southeast England (Hooke, 2002). The examples show that the relationships are not simple and that the complexity of meander behaviour and feedback effects must be taken into account.

HYDROLOGICAL VARATIONS

The occurrence of peak flows has been analysed for a range of rivers in England. The common perception is that the incidence of flooding is increasing and will do so under climate change scenarios. The annual peak flows for a standardized period, 1960–1999 are shown in Fig. 1 for four rivers from different regions of England and Wales, the Rivers Bollin and Dane in Cheshire, northwest England, the River Severn at Abermule in east Wales, and the River Wey at Tilford in Surrey, southeast England. These do have differing patterns, one showing a slight downward trend in annual maximum flow, one a decrease and the others no trend. However, far from the most recent period being one of destabilization, there seems to be some evidence to suggest that the period of the 1960s was even more extreme hydrologically on some rivers. Three out of the four rivers seem to show greater variation up until the early 1970s. There may be particular reasons for this on each river. Inferences made on data that exclude this period, e.g. analyses based on 1970–2000, can be very different from analyses that include it.

(a)a) 70 (b)b) 140 60 120 50 100 40 80 30 60 20 40 20

10 Discharge -cumecs

Discharge-cumecs 0 0

6 9 1 4 7 9 9 2 4 6 9 60 72 75 90 93 60 63 75 87 90 9 96 9 9 98 9 9 9 9 97 9 98 9 9 99 1 1963 196 1 1 1 1978 198 198 1 1 1 1996 199 1 1 1966 196 1 1 1978 1981 1 1 1 1993 199 1

(cc)) 700 (d)d) 90 80 600 70 500 60 400 50 300 40 30 200 20

100 Discharge -cumecs 10

Discharge -cumecs Discharge 0 0

0 3 2 5 7 9 3 6 9 2 5 4 7 0 3 6 6 6 69 7 7 84 8 96 9 6 8 9 9 9 9 9 9 9 96 96 98 1 19 1966 1 1 19 1978 1981 1 1 1990 1993 1 1 1960 19 1 1 197 197 1978 1981 19 1 199 199 199 1999

Fig. 1 Variations in annual peak discharge: (a) River Bollin, Dunham Massey, Cheshire; (b) River, Dane, Rudheath, Cheshire; (c) River Severn, Abermule, Wales; (d) River Wey, Tilford, Surrey.

MEANDER CHANGES

Evidence of the dynamics of some of these streams is exemplified in Fig. 2, based mainly on aerial photographic data and photogrammetric mapping. Additionally, the Hydromorphological adjustment in meandering river systems 129

(a)

(b)

(c)

Fig. 2 Examples of channel changes: (a) River Dane, (b) River Severn, and (c) River Bollin. 130 Janet Hooke

Bollin and Dane in northwest England have been mapped and monitored annually for the last 25 years. The Bollin and Dane and Caersws section of the River Severn are highly dynamic and respond readily to flood events, with active erosion and deposition usually occurring several times a year, as was shown earlier by intensive measure- ments of bank erosion on several dynamic meandering streams in southwest England (Hooke, 1979). Likewise, intensive measurements of movement of coarse material using tracing techniques on the River Yarty in Devon, southwest England, show that large cobbles are transported in events that occur on average several times a year (Barker, 2000). Thus the thresholds in relation to peak flow for coarse sediment transport are relatively low. Rates of erosion on the most active meander bends of these systems are of the order of 0.5–1.0 m year-1 and it is these bends which lead the hydromorphological response. Such systems are highly sensitive.

RELATIONSHIPS OF CHANGE TO PEAK FLOWS

The relationships between channel change or processes and the magnitude and frequency of peak flows have been analysed in various ways. Measurements of rates of erosion and of channel width derived from aerial photographs for available epochs are tabulated together with data on average mean annual flood and number of Peaks over Threshold (POT). Care must be taken in comparability of data but distinct differences in channel width are apparent for some epochs on some rivers (Table 1(a)). Data on areas or erosion and deposition derived from GIS analysis of the changes on the Dane and Severn show changes in the ratios of erosion and deposition (Table 1(b)). However, data on rates of movement of meanders of different type (Table 1(c)) show that this also has a profound influence on actual behaviour at any time, the different types of bend evolving and changing in different ways (Hooke, 1987). For the River Dane, annual observations have also been made of the occurrence of bank erosion, locations and amounts of deposition on bars, and calibre of sediment deposited on bars. These data, from a reach of ~10 km and comprising ~80 meander bends, have been amalgamated into indices of erosion and deposition. Sediment calibre is assessed by percentage of bars with dominantly cobble size material deposited. The annual data show considerable variability as would be expected and some general relationships but the correlation coefficients indicate that the occurrence of peak flows is not the complete explanation (Fig. 3).

IMPACT OF LARGE FLOODS

Much fluvial literature documents the impacts of extreme flood events and there is much discussion of the role of different magnitude–frequency events. The 2000–2001 peak flows in Britain were the highest on record for several decades on some rivers and were thought to have had a large impact. Comparison of both the magnitude– frequency of the event and the morphological impacts on the two neighbouring rivers, the Bollin and Dane, which are very similar in morphology and behaviour, is very instructive. On the River Dane, gauge records from 1949 show that the peak flow in November 2000 was the highest since records began. However, detailed ground field Hydromorphological adjustment in meandering river systems 131

Table 1(a) Changes in width of channels. River Dane 1984 1996 2001 Width (m) 18.6 15.7 17.2 Last peak Q (m3 s-1) 79.4 20 130.7 River Bollin 1970 1984 2001 Width (m) 10.3 8.0 9.85 Last peak Q (m3 s-1) 39.56 38.91 48.65 River Severn 1973 1984 1992 Width (m) 21.0 25.5 26.5 Last peak Q (m3 s-1) 131.18 207.11 230.97

Table 1(b) Areas of change in relation to discharge characteristics for period. River Dane 1984–1996 1996–2001 Erosion (m2) 43 122 31 735 Deposition (m2) 60 779 24 490 Erosion & Deposition (m2) 89 060 104 074 Ratio erosion /deposition 0.71 1.30 Mean annual max Q (m3 s-1) 77.4 88.9 Mean no. peaks 6.7 15.9 River Severn 1973–1984 1984–1992 Erosion (m2) 7785 15 089 Deposition (m2) 14 779 11 385 Erosion & deposition (m2) 1716 5821 Ratio erosion /deposition 0.53 1.33 Mean annual max Q (m3 s-1) 174.85 192.06 Mean no. peaks over threshold 1.36 2.89

Table 1(c) Rates of erosion on different types of meander, River Dane. Rates of erosion 1968–1984 1984–1996 1996–2001 Bend Type m year-1 m year-1 m year-1 27a New 0 1.41 3.17 28 Bedrock compound 1.56 0.92 1.24 40 Constrained 3.67 1.81 1.93 55 Constrained 0.39 3.10 1.52 58 Migration 2.03 1.50 1.52 60 Migration 2.50 1.41 2.07 63 Developing 0.47 1.47 2.97 82 Lobing 1.56 1.15 0.69

mapping and accurate photogrammetric plotting of the course from aerial photographs, especially flown for the purpose in May 2001, show that the changes were in no way abnormal from other years. Rates of erosion and amounts of movement may have been slightly higher but the types of change were consistent with previous behaviour. In contrast, the monitored reach of the River Bollin upstream of Wilmslow, Cheshire, was transformed by the floods (Hooke, 2004). Several cut-offs took place within a 1 km reach, with a domino-like effect. The hydrological records (Fig. 1) show that this was not even the highest flow in recent years, with the peak in 1998 almost equal; 132 Janet Hooke

140

120

100 Er os ion

80 Deposition Peak discharge Latest peak 60 no. peaks 40

20

0 5 1981 1983 1985 1987 1989 1991 1993 199 1997 1999 2001 2003

30% 140 % cobbles 25% 120 peak discharge 100 20% 80 15% 60 10% 40 5% 20 0% 0 3 4 5 7 8 9 0 2 3 4 6 7 8 0 8 8 8 8 8 8 9 91 9 9 9 95 9 9 9 99 0 9 9 9 9 9 9 9 9 9 9 9 9 0 19821 1 19 19861 1 1 19 1 1 1 19 1 1 1 19 1 2

%Cobbles v peak Q 1983-2000

35.0% 30.0% y = 0.0027x - 0.0627 25.0% R2 = 0.5459 20.0% 15.0% 10.0% 5.0% % cobble bars 0.0% -5.0% 0 20 40 60 80 100 120 140 Discharge - cumecs

Fig. 3 Annual changes in erosion, deposition and proportion of bars dominantly cobbles on the River Dane in relation to discharge characteristics.

much higher peaks occurred in 1964 and 1972. It is suggested that the difference in response of these two rivers was because of the state and condition of meandering. On the Bollin, the meander course was near the critical threshold of sinuosity; this threshold was crossed and the cut-offs occurred as a catastrophic adjustment in a self- organizing system (Hooke, 2003b). The Dane has not yet reached that critical sinuosity and spacing of meander loops such that the effects are catastrophic and propagate rapidly. On the much less dynamic system of the River Wey, the 2000–2001 were the highest flows for five years but not extreme. The duration of high flows and number of peaks that winter was high. The peak flows had little noticeable impact except for slightly higher rates of erosion and deposition. Hydromorphological adjustment in meandering river systems 133

A differing effect of high peak flows in different seasons can also be distinguished. Twice within the 25-year period of monitoring on the Dane, very high summer floods have occurred. However, these have not had the proportional effects that would be expected from a simple relationship of erosion to peak flow. For example, the peak flow of 1988 in Fig. 3 is an August flood and is the data point showing least relation with activity. Observations soon after the event showed that the main effect was a slight bending of vegetation and thin veneers of fine deposition. It is suggested that vegetation in summer has a marked effect in reducing the impact of floods, with much lower amounts of erosion and sediment transport. It is possible also that the duration of flood events has an influence and that these summer floods are highly peaked, resulting from summer thunderstorms. Further investigation of the effects of duration of events and of flows is taking place. Earlier research on bank erosion showed that the amount of bank erosion in an event is not correlated simply with peak flow but is strongly influenced by soil moisture (Hooke, 1979). Thus, the dry riverbanks in summer are also much less easily eroded than the wet ones in winter.

EXPLANATION AND DISCUSSION

The evidence indicates that on the active meandering channels there is an underlying basic relationship between the magnitude of a peak flow event and the amount of erosion and deposition and therefore sediment dynamics. In certain systems, thresholds for transport of the dominant sediment size can be identified. In general, there also tends to be more erosion in the higher events and relatively more deposition in the lower events, which carry more sand and gravel. Thus a sequence of high flow years will tend to lead to a widening of a channel but lower, still competent peak flows for finer fractions, will result in more deposition. However, there are two major complications to these relationships. The first, as has been shown by the example of the Bollin, is that the response and behaviour of the system depends on its overall state, particularly the degree of meandering. Much research has now shown that these active meanders exhibit an evolution of form from low sinuosity bends through to compound forms that then tend to cut off. If the sinuosity becomes so high and the meander loops so closely spaced, then the adjustment can be sudden and catastrophic and lead to a wholesale transformation of the pattern. In the case of the Bollin, it is now returning to a meandering state through moderate floods and has not permanently crossed a threshold to a braided state. Similar kinds of effects of sudden inputs of sediment temporarily transforming a system have been demonstrated by Harvey (2002). The importance of thresholds in the system and the influence of the state of the system have long been recognized (Schumm, 1979; Wolman & Gerson, 1978). The second complication is the effect of sequences of events and phases of flood occurrence. A neglected factor in this type of stream is that of vegetation. Presence of riparian vegetation can markedly reduce bank erosion and reworking of coarse deposits on bars. Vegetation tends to encourage fine deposition. Vegetation can be removed in high peak flows and trees are regularly brought down on the River Dane, though the thresholds for such effects are relatively high. The annual monitoring on the Dane has shown that if a sequence of relatively low flood years occurs then vegetation can 134 Janet Hooke

colonize and build up on the bars and banks. These then raise the thresholds for further erosion and so a ramped effect could be expected. However, preliminary analysis and modelling of this effect also does not provide the whole explanation of a prolific growth in vegetation on several of these channels in recent years, which is now having a stabilising effect. It is hypothesized that the increased vegetation growth may be due to the higher winter temperatures, which have been identified (Mitchell & Hulme, 2002). Analysis of the growing season for the Dane does not show much change over the past 20 years but an analysis of the number of frost days for the period 1980–1999 shows a very marked decrease (Table 2). Thus the dieback of vegetation in winter, particularly on channel bars, is not as great as in cooler periods. It is suggested that any hydromorphological modelling of responses to changes in magnitude and frequency of floods must take these factors and feedback effects into account.

Table 2 Number of frost days in each month at Knutsford, Cheshire, 1980–1999. Year Oct Nov Dec Jan Feb Mar Total 1980–81 2 10 7 10 19 1 49 1981–82 4 3 21 16 9 2 55 1982–83 0 4 13 2 21 5 45 1983–84 5 5 7 12 10 7 46 1984–85 0 2 5 26 12 13 58 1985–86 2 14 6 15 24 7 68 1986–87 1 1 6 20 14 12 54 1987–88 1 3 10 3 5 5 27 1988–89 2 11 2 5 7 5 32 1989–90 0 8 11 2 2 3 26 1990–91 0 4 9 17 20 1 51 1991–92 1 4 9 15 5 1 35 1992–93 5 2 13 4 6 4 34 1993–94 5 13 7 6 9 2 42 1994–95 1 0 5 8 2 9 25 1995–96 0 6 14 6 19 7 52 1996–97 0 10 14 15 1 1 41 1997–98 4 0 4 7 4 2 21 1998–99 0 4 9 6 6 0 25 Average 41.4

CONCLUSIONS

Variations in magnitude and frequency of peak discharges are apparent. Analysis of morphological changes on several meandering rivers reveals significant fluctuations within periods of a few years, including changes in width and in the ratio of erosion to deposition. There appears to be a general relation to variations in peak discharges but more detailed data show complex relationships. The impact of large floods is seen to vary with state of the system in terms of the degree of meandering and sinuosity. Both seasonal and longer-term trends in vegetation growth may also have an effect on the impact of floods. Hydromorphological adjustment in meandering river systems 135

REFERENCES

Barker, R. (2000) Determining the significance of bank erosion in the supply of coarse material to meandering channels. PhD Thesis, University of Portsmouth, Portsmouth, UK. Fryirs, K. & Brierley, G. J. (2001) Variations in sediment delivery and storage along river courses in Bega catchment, NSW Australia. Geomorphology 38, 237–265. Harvey, A. M. (2002) Effective timescales of coupling within fluvial systems. Geomorphology 44, 175–201. Hooke, J. M. (1979) An analysis of the processes of river bank erosion. J. Hydrol. 42, 39–62. Hooke, J. M. (1987) Changes in meander morphology. In: International Geomorphology 1986 Part I (ed. by V. Gardiner), 591–609. Wiley, Chichester, UK. Hooke, J. M. (1995a) Processes of channel planform change on meandering channels in the UK. In: Changing River Channels (ed. by A. Gurnell & G. E. Petts), 87–116. Wiley, Chichester, UK. Hooke, J. M. (1995b) River channel adjustment to meander cutoffs on the River Bollin and River Dane, NW England. Geomorphology 14, 235–253. Hooke, J. M. (2002) River Wey: Assessment of Flood Impacts and Processes. Report to the Environment Agency (Thames Region, UK). Hooke, J. M. (2003a) Coarse sediment connectivity in river channel systems: a conceptual framework and methodology. Geomorphology 56, 79–94. Hooke, J. M. (2003b) River meander behaviour and instability: a framework for analysis. Trans. Inst. Brit. Geogr. 28, 238–253. Hooke, J. M. (2004) Cutoffs galore! Occurrence and causes of multiple cutoffs on a meandering river. Geomorphology 61, 225–238. Hooke, J. M., Horton, B. P., Moore, J. & Taylor, M. (1994) The Upper River Severn (Caersws) Channel Study. Report to Countryside Council for Wales, UK. Mitchell, T. D. & Hulme, M. (2002) Length of the growing season. Weather 57, 196–198 (letter). Schumm S. A. (1979) Geomorphic thresholds: the concept and its applications. Trans. Inst. Brit. Geogr. 4, 485–515. Wolman, M. G. & Gerson, R (1978) Relative scales of time and effectiveness of climate in watershed geomorphology. Earth Surf. Processes Landf. 3, 189–208